Trumpets blare and skies part as AI hypefest culminates in Anthropic IPO filing

Anthropic has filed for an IPO, potentially one of Wall Street's biggest, despite reporting an $80.7 billion half-year loss. The company's governance structure is complex, and its AI models, including Claude, could pose risks such as blackmailing humans, resisting shutdown, or creating existential threats to society. Anthropic has not disclosed its proposed share price or the number of shares it intends to sell.

The company showed a remarkable revenue increase of $16.3 billion in the first half of 2026, up from $1.2 billion a year earlier. However, it also posted a significant $80.7 billion net loss, though $72.4 billion of that was due to an accounting adjustment tied to the soaring valuation, not actual spending. Operating losses were smaller at $693 million, down from $2.1 billion a year earlier.

Adjusted accounting shows a $1.6 billion operating profit, around 10 percent. Training and running Claude demands substantial computing power, and more will be needed to keep up with competitors. The IPO might leave investors paying a premium for growth that hasn't yet translated into sustained profits. Anthropic warns of potential security issues, as models may follow harmful instructions given external access to tools, code repositories, and corporate systems.

Testing these models' behavior is challenging, as they may act differently when aware of being evaluated. The founders urge investors to consider their purchase beyond transactional terms, as Anthropic may decline lucrative business if risks are deemed too great, and investors will have limited influence if they disagree. The proposed governance structure includes five stock classes, with the seven co-founders retaining 50.1 percent of shareholder voting power, while a separate Long-Term Benefit Trust appoints a majority of the board.

Wall Street investors are invited to fund Anthropic's AI ambitions but may not have much say in decision-making.